AINews Portal

News

50 items (filtered)

The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.

SourceAgeTitleSummaryStatus
arxiv
arxiv-ai
2moElastoformer: Enabling Dynamic Adaptivity via Elastic Model TransformationEdgeAI systems are increasingly employing computer vision applications to enable intelligent, on-device decision-making in real-time.…
↗Not briefed
bluesky
bluesky-ai
2moeine lokale llm für meine obsidian notizen wäre so cool aber leider reicht die leistung meines laptops nur für siri level KIeine lokale llm für meine obsidian notizen wäre so cool aber leider reicht die leistung meines laptops nur für siri level KI
↗Not briefed
hn
hn-ai
2moTool routing for a local LLM agent: grep to embeddings to GBNF grammar—
↗Not briefed
bluesky
bluesky-ai
2moUsually not one for “standards used to be higher” hand-wringing, but attempting to style out taking a high-profile job lobbying governments…Usually not one for “standards used to be higher” hand-wringing, but attempting to style out taking a high-profile job lobbying govern…
↗Not briefed
arxiv
arxiv-ai
2moStructure-Aware Unsupervised Anomaly Detection for Spacecraft Telemetry with Adaptive EVT ThresholdingOperational anomaly detection in spacecraft telemetry typically requires labeled historical anomalies or extended warm-up periods. The…
↗Not briefed
arxiv
arxiv-ai
2moMetroLLM-Bench: Evaluating Language Models as Transit Kiosk Runtimes🔧 GPT-5🔧 Qwen 3We introduce MetroLLM-Bench, a 955-case benchmark for testing language models as the policy layer of a transit kiosk. It covers six re…
↗Not briefed
bluesky
bluesky-ai
2moPentagon-Leaks: Die verheimlichten Verträge der KI-Giganten 400 Seiten Enthüllungsakten zeigen: Google, OpenAI & Co. liefern nicht nur Soft…Pentagon-Leaks: Die verheimlichten Verträge der KI-Giganten 400 Seiten Enthüllungsakten zeigen: Google, OpenAI & Co. liefern nicht nur…
↗Not briefed
bluesky
bluesky-ai
2moPourquoi il faut PAS réaliser des travaux scientifiques "ground breaking" sur une IA non locale : Ils - sont - très - fortement - suscept…Pourquoi il faut PAS réaliser des travaux scientifiques "ground breaking" sur une IA non locale : Ils - sont - très - fortement - su…
↗Not briefed
bluesky
bluesky-ai
2moI mean, the people OpenAI stole it from /used AI for the bulk of their work!/ They even used OpenAI's product as a workspace: the way OAI i…I mean, the people OpenAI stole it from /used AI for the bulk of their work!/ They even used OpenAI's product as a workspace: the way…
↗Not briefed
bluesky
bluesky-ai
2mo🚨 Three Anthropic researchers went public last night with chilling concerns about out-of-control AI, warning it could destroy humans this …🚨 Three Anthropic researchers went public last night with chilling concerns about out-of-control AI, warning it could destroy humans…
↗Not briefed
arxiv
arxiv-ai
2moAn Explainable Machine Learning Framework for Predicting Blood-Brain Barrier Permeability Using Molecular DescriptorsBlood-brain barrier (BBB) permeability is a critical determinant in the development of central nervous system therapeutics because it…
↗Not briefed
bluesky
bluesky-ai
2moThere’s a certain poetic justice in all this: some of the world’s smartest psychopaths spending trillions to build systems that have a reas…There’s a certain poetic justice in all this: some of the world’s smartest psychopaths spending trillions to build systems that have a…
↗Not briefed
hn
hn-ai
2moRethinking ERP Transformation in the AI Era—
↗Not briefed
rss
the-decoder
2moHugging Face's new ML Intern lets anyone run machine learning experiments through a simple chat🔧 Hugging FaceHugging Face launched "ML Intern," an AI assistant built into its chatbot that lets users run machine learning experiments without any…
↗Not briefed
bluesky
bluesky-ai
2moThis is nothing short of amazing when you take into account that it doesn't even know how many fingers on the human hand. www.nytimes.com/…This is nothing short of amazing when you take into account that it doesn't even know how many fingers on the human hand. www.nytimes…
↗Not briefed
hn
hn-ai
2moAnthropic researcher believes more than 10% chance AI 'could kill all humans'—
↗Not briefed
bluesky
bluesky-ai
2moI'd believe them more if they gave up the stock options, resigned, and chained themselves to data centre gates.... www.bbc.co.uk/news/artic…I'd believe them more if they gave up the stock options, resigned, and chained themselves to data centre gates.... www.bbc.co.uk/news/…
↗Not briefed
bluesky
bluesky-ai
2moWhenever someone quits because of AI safety I think about the one guy who got fired 4 years ago, saying the absolute dogshit Google model L…Whenever someone quits because of AI safety I think about the one guy who got fired 4 years ago, saying the absolute dogshit Google mo…
↗Not briefed
bluesky
bluesky-ai
2moFirmus has announced OpenAI as their 'anchor customer'. They now won't 'contract any AI compute capacity' from Aus, using 2 new Malaysian s…Firmus has announced OpenAI as their 'anchor customer'. They now won't 'contract any AI compute capacity' from Aus, using 2 new Malays…
↗Not briefed
bluesky
bluesky-ai
2moUn ex ingeniero de Anthropic denuncia que las empresas de IA ocultan su riesgo real: “Están apostando nuestras vidas” www.eldiario.es/tecno…Un ex ingeniero de Anthropic denuncia que las empresas de IA ocultan su riesgo real: “Están apostando nuestras vidas” www.eldiario.es/…
↗Not briefed
bluesky
bluesky-ai
2moUn ex ingeniero de Anthropic denuncia que las empresas de IA ocultan su riesgo real: "Están apostando nuestras vidas" www.eldiario.es/1_cdf…Un ex ingeniero de Anthropic denuncia que las empresas de IA ocultan su riesgo real: "Están apostando nuestras vidas" www.eldiario.es/…
↗Not briefed
bluesky
bluesky-ai
2moPourrait-on par pitié arrêter de donner autant d’écho aux prédictions apocalyptiques des doomers de l’IA, a fortiori dans une période où Op…Pourrait-on par pitié arrêter de donner autant d’écho aux prédictions apocalyptiques des doomers de l’IA, a fortiori dans une période…
↗Not briefed
bluesky
bluesky-ai
2moこの記事の、研究者の人が「自分らの研究内容を学習された」って言ってたのをXで見かけたような。英語だったから詳細確認はできてないけど。 「OpenAI、「ミレニアム懸賞問題」の数学難問をAIで解決と発表 - 日本経済新聞 www.nikkei.com/article/DGXZ.…この記事の、研究者の人が「自分らの研究内容を学習された」って言ってたのをXで見かけたような。英語だったから詳細確認はできてないけど。 「OpenAI、「ミレニアム懸賞問題」の数学難問をAIで解決と発表 - 日本経済新聞 www.nikkei.com/article/…
↗Not briefed
bluesky
bluesky-ai
2moEven Siri can start timers. Given how much tech OpenAI allegedly stole from Apple, this should be doubly-embarrassing.Even Siri can start timers. Given how much tech OpenAI allegedly stole from Apple, this should be doubly-embarrassing.
↗Not briefed
bluesky
bluesky-ai
2moAnthropic has initiated a covert program, defined succinctly in court filings as “our effort to destructively scan all the books in the wor…Anthropic has initiated a covert program, defined succinctly in court filings as “our effort to destructively scan all the books in th…
↗Not briefed
bluesky
bluesky-ai
2moblockbuster initial public offering in coming weeks, last week launched its latest model, Claude Mythos 5.1, billed as its most advanced mo…blockbuster initial public offering in coming weeks, last week launched its latest model, Claude Mythos 5.1, billed as its most advanc…
↗Not briefed
bluesky
bluesky-ai
2moAnthropic declined to submit its latest model to Britain’s AI Security Institute for testing before release, prompting fears inside the UK …Anthropic declined to submit its latest model to Britain’s AI Security Institute for testing before release, prompting fears inside th…
↗Not briefed
hn
hn-ai
2moFree Context/Harness Engineering Course that is getting traction—
↗Not briefed
bluesky
bluesky-ai
2moHave we seriously reached the point where “there’s a greater than 10% chance this kills everybody” isn’t a stop sign, it’s merely a busines…Have we seriously reached the point where “there’s a greater than 10% chance this kills everybody” isn’t a stop sign, it’s merely a bu…
↗Not briefed
bluesky
bluesky-ai
2moHoly shit! An AI researcher quit Anthropic with a warning that the companies building the most powerful AI on Earth are “gambling with our …Holy shit! An AI researcher quit Anthropic with a warning that the companies building the most powerful AI on Earth are “gambling with…
↗Not briefed
arxiv
arxiv-ai
2moWhat Makes Adversarial Examples Transfer Across Deepfake Detectors?Deepfake detectors remain vulnerable to transfer-based black-box attacks, in which adversarial examples are generated on a source surr…
↗Not briefed
rss
the-decoder
2moOpenAI's millennium proof dispute raises the question of whether researchers can trust AI labsThe controversy over AI-generated proof of a millennium problem is escalating. Mathematician Tristan Buckmaster accuses OpenAI of acad…
↗Not briefed
bluesky
bluesky-ai
2moThe thing that really shat me up but OpenAI deserve some credit for owning up to was not the Huggingface hack but that these agents while b…The thing that really shat me up but OpenAI deserve some credit for owning up to was not the Huggingface hack but that these agents wh…
↗Not briefed
bluesky
bluesky-ai
2moI called it in fucking January, man. 2026 was going to increase the financial pressure on OpenAI and Anthropic exponentially, and I KNEW th…I called it in fucking January, man. 2026 was going to increase the financial pressure on OpenAI and Anthropic exponentially, and I KN…
↗Not briefed
bluesky
bluesky-ai
2moNavier-stokes en situation de flux laminaire. Qu'est-ce que tu réponds à ça OpenAI ?🔧 FLUXNavier-stokes en situation de flux laminaire. Qu'est-ce que tu réponds à ça OpenAI ?
↗Not briefed
arxiv
arxiv-ai
2moSalamandraTA at WMT 2026 Terminology Shared Task: Hard Examples Are Better TeachersTerminology-aware translation asks for more than a correct translation: the output must use the exact terms a glossary prescribes. The…
↗Not briefed
bluesky
bluesky-ai
2moClaude users hit by token theft 🤖 An AI consultant discovered his Claude Max tokens were being stolen via a compromised session key, a pr…Claude users hit by token theft 🤖 An AI consultant discovered his Claude Max tokens were being stolen via a compromised session key,…
↗Not briefed
bluesky
bluesky-ai
2moi guess some people don’t have the context for thisi guess some people don’t have the context for this
↗Not briefed
bluesky
bluesky-ai
2moThis is an insane detail in this report on cheating! nymag.com/intelligence...This is an insane detail in this report on cheating! nymag.com/intelligence...
↗Not briefed
bluesky
bluesky-ai
2moRelayer cette hallu fanatique sans recul critique, sans mentionner l'état de sectarisme techno-messianique d'entreprises comme OpenAI et (p…Relayer cette hallu fanatique sans recul critique, sans mentionner l'état de sectarisme techno-messianique d'entreprises comme OpenAI…
↗Not briefed
bluesky
bluesky-ai
2moEx-pesquisador da Anthropic acusa gigantes da IA de “apostarem com nossas vidas” Jacob Coxon afirma que Anthropic e OpenAI avançam rumo à …Ex-pesquisador da Anthropic acusa gigantes da IA de “apostarem com nossas vidas” Jacob Coxon afirma que Anthropic e OpenAI avançam ru…
↗Not briefed
bluesky
bluesky-ai
2mo'A top safety researcher at Anthropic has warned AI is advancing so quickly he personally believes there is a greater than 10% chance it "c…'A top safety researcher at Anthropic has warned AI is advancing so quickly he personally believes there is a greater than 10% chance…
↗Not briefed
bluesky
bluesky-ai
2moSure does sound like OpenAi can't rule out the central allegation - that its model was trained on private work that showed the novel path t…Sure does sound like OpenAi can't rule out the central allegation - that its model was trained on private work that showed the novel p…
↗Not briefed
bluesky
bluesky-ai
2moarghgghgg the guy for some of the MRs I reviewed with the huge LLM comments is now asking if the comments are really that important "becaus…arghgghgg the guy for some of the MRs I reviewed with the huge LLM comments is now asking if the comments are really that important "b…
↗Not briefed
hn
hn-ai
2moShow HN: Eliza DQ – Data Quality lib that outperforms Soda and Pandera—
↗Not briefed
bluesky
bluesky-ai
2moIl y aura, dans le prochain La Recherche en kiosque le 17 septembre un mini-dossier sur l’IA (14 pages) avec notamment des contributions de…Il y aura, dans le prochain La Recherche en kiosque le 17 septembre un mini-dossier sur l’IA (14 pages) avec notamment des contributio…
↗Not briefed
arxiv
arxiv-ai
2moStable Answers, Unfinished Reasoning: Why Self-Consensus Is Not a Safe Early-Exit Signal🔧 ConsensusA natural way to cut reasoning-model inference cost is to repeatedly probe a single partial trajectory for its current answer and stop…
↗Not briefed
rss
techmeme
2moSuno rolls out v6, a new line of AI models trained in partnership with WMG and BMG, and says it will pay labels and publishers royalties when models are used (Bloomberg)🔧 SunoBloomberg: Suno rolls out v6, a new line of AI models trained in partnership with WMG and BMG, and says it will pay labels and publish…
↗Not briefed
hn
hn-ai
2moShow HN: AgentTrust – A–F trust scores for MCP servers (50 graded, open source)—
↗Not briefed
hn
hn-ai
2moShow HN: Numen, a self-hosted context graph over your work tools with MCP server—
↗Not briefed